Re: Raspberry Pi 3B+ and pitiful network speeds

From: Denis Ovsienko <denis_at_ovsienko.info>
Date: Wed, 23 Jun 2021 12:47:41 +0100
On Wed, 23 Jun 2021 12:30:29 +1000
MJ <mafsys1234_at_gmail.com> wrote:

> I will likely abandon using FreeBSD, but I should probably raise a
> bug report for this? I am not sure of the process for this or whether
> it's even wanted. That is, does FreeBSD care enough about little
> appliances like the Raspberry Pis to spend time fixing an obviously
> broken network driver/stack?

It seems to me that overall support is likely to improve in some
future. Little boards are good vehicles for ramping newcomers up into
the ARM space, so they do receive developer attention. That said,
there's a lot of hard work involved in getting an operating system to
support a particular board fully, and in some parts of the world some
boards are more available than others. RPI is the type most available
to me, and it is very helpful that FreeBSD provides a ready to run RPI
image for download, and treats ARM as a Tier 1 architecture.

Regarding your best next steps, it might help to consider two aspects of
"works": functional-wise and performance-wise. Clearly, in your case
there is a performance issue, and the best you can do is raise a
detailed bug report.

Functional-wise the fact is, some hardware features (GPIO, audio,
accelerated video, I2C, WiFi, Bluetooth, etc) are not supported well or
at all by particular releases of particular BSD systems on particular
boards (cannot remember specifics, and it is changing all the time). So
if your use case involves anything beyond USB, Ethernet (and even SD
card, as it was with OpenBSD until recently!), you would have to
double-check before switching to another board or another OS.

With that in mind, if your project's functional requirements allow to
switch to RPI4B, it might be another good thing to do whilst your
RPI3B+ bug report is waiting to be resolved. There's plenty of churn in
RPI4B code and it looks unlikely to fully settle anytime soon (FreeBSD
is not unique in this regard), but Ethernet works OK. I have just done
a quick one-way TCP throughput tests on FreeBSD/RPI4B and it resulted
in 87MB/s in either direction. That's an order of magnitude more than
what you are observing with RPI3B+.

Once again, an even simpler way out of the performance issue might be
through a USB Ethernet interface that happens to "just work", whatever
the underlying reasons are.

-- 
    Denis Ovsienko
Received on Wed Jun 23 2021 - 11:47:41 UTC

Original text of this message